Research On Resource Optimization Problems In Cognitive Wireless Networks

With the rapid development of wireless communication technologies and the increasing broadband communication demands, the shortage of available spectrum is becoming an outstanding problem. With the aid of some advanced wireless communication technologies, such as link adaptation and multiple antennas, the efficiency of spectrum usage in existing networks is extremely improved. As shown in some research reports, however, the average proportion of the total available spectrum used by people is only 2%-6%. The reason is that the static policy management policy entails higyly inefficiency resource utilization. Under the above background, the cognitive radio technology is proposed. Cognitive radio is an intelligent radio device based on software defined radio, enabling the communication requirements of cognitive users and the high spectrum utilization with various intelligent modules.Currently, cognitive wireless network, based on cognitive radio, becomes a hot topic in wireless communication research. Related research are undergoing by research organizations and scholars from universities all over the world. Nevertheless, the architecture, protocols and function modules of the cognitive wireless network are also in the first stage. In this paper, our researches aim to solve the resource optimization problems in cognitive wireless network, and focus on dynamic learning, MAC spectrum access, and research on existing networks with virtual cognitive modules. The researches have meaningful value for the development of cognitive radio networks.The work in this dissertation is concluded as following aspects. Firstly, the research progress of cognitive wireless network is dicussed detailedly, and the characteristic and technology challeges are analyzed.Secondly, in chapter 3, we defined the environmental dynamics classification of cognitive wireless networks, which are respectively caused by dynamic channel state and competition among cognitive users. In addition, we designed three dynamic learning mechanisms for different network models. The first mechanism is designed as an auction algorithm that achieves the dynamic power allocation for spectrum sharing. We only consider the second dynamics in the first mechanism, and also analyse the perforemance on users' long-term utility by dynamic learning and competing with other users. Under the open sharing model, the second mechanism is proposed based on cooperative game theory, only relied on the second dynamics. At last, according to control and decision theory, we defined a new valuation for cognitive users, considering the long-term energy efficiency. Based on reinforcement learning, the third mechanism is proposed to solve the robust transmission for cognitive users, considering the general dynamics. All the three mechanisms can be all distributed algorithms, and aims to improve the energy efficiency for cognitive users while reducing the interaction cost.Thirdly, chapter 4 discussed the MAC spectrum access mechanisms in cognitive wireless networks. In the scenario coexisted with tranditional and cognitive wireless networks, the main access scheme for cognitive users are spectrum sharing, opportunistic spectrum access and spectrum market, with the assumption that cognitive users must be always "waiting"-until the spectrum opportunities appear. In this paper, we model the channel usage of liscensed users, and propose a new MAC spectrum access mechanism relied on cooperative communication. In existing wireless networks (such as celluar network), there are liscensed users with bad communication links that guarantee their QoS. Therefore, cognitive users can utilize this character to perform as relays, build coorperative transmition with licensed users, and obtain the spectrum resource. Based on this mechanism, we propose three available coorperative transmition schemes, and respectively calculate the expectation of spectrum opportunity and consumed energy for spectrum access. The proposed mechanism realized the automation for cognitive users, and obtained more opportunities than tranditional menchanisms.Fouthly, in chapter 5 we studied the performance on exsiting wireless networks with virtual cognitve modules. In this chapter, we firstly studied the subcarrier allocation in multi-cell OFDMA system with coordinanted multipoint transmission, which is discussed in 3GPP LTE-Advanced to avoid the inter-cell interference and improve the performance of cell-edge users. We model the subcarrier allocation problem as a dynamic integer programming, and design a suboptimal algorithm with an adjusted frequency reuse factor. Numerical results showed that the performance of the proposed algorithm is better enough. At last, we assume a virtual cognitive module deployed in each base station, and transfer the above problem into a dynamic frequency programming. Through designing a learning algorithm for the virtual cognitive module, the performance of the dynamic frequency programming is proved to be closed to the former algorithms.Finally, a conclusion is addressed for the dissertation, and valuable research directions in the future are presented.